Dynamic polarization-beam steering anti-interference method and system based on intelligent metasurface

Abstract

The present invention provides a dynamic polarization-beam control and anti-interference method and system based on intelligent metasurfaces, which relates to the field of wireless communication technology, including an intelligent metasurface array, a controller, a channel sensing module and a joint optimization module. By real-time sensing the channel state and interference characteristics in the wireless environment, the polarization state and phase configuration of the intelligent metasurface unit are dynamically adjusted to achieve collaborative optimization of the polarization domain and the spatial domain, forming an enhanced propagation path for the desired user, while suppressing the interference source, improving the system's anti-interference capability and communication quality. The present invention adopts a hybrid optimization strategy that combines deep reinforcement learning with model-driven, which reduces computational complexity and achieves rapid response to dynamic interference environments. At the same time, it improves spectrum efficiency and energy efficiency through polarization-beam joint optimization. Through collaborative optimization of the polarization domain and the spatial domain, an adaptive response to complex interference environments is achieved, while reducing energy consumption and implementation costs.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technology, and in particular to a dynamic polarization-beam steering anti-interference method and system based on intelligent metasurface. Background Art

[0002] Intelligent Reflecting Surfaces (IRS), a key enabling technology for 6G wireless communications, have gradually attracted widespread attention from researchers. Comprised of a large number of programmable electromagnetic units, IRSs can actively control the reflection, refraction, or scattering of electromagnetic waves by adjusting the electromagnetic properties of each unit, thereby reshaping the propagation environment for wireless signals. Currently, research on intelligent metasurfaces focuses on the following areas:

[0003] 1. Beamforming: Through phase manipulation of intelligent metasurface units, beam directivity is optimized, thereby enhancing signal strength in specific directions. This method primarily optimizes signal propagation in the spatial domain, avoiding interference sources to a certain extent and improving system signal quality.

[0004] 2. Polarization Control Technology: Utilizing polarization properties to enhance the anti-interference capabilities of communication systems. Through polarization diversity coding or polarization control, signal transmission can be independent, effectively preventing interference from other sources. Some research has combined polarization control technology with intelligent metasurfaces, further improving channel capacity and system reliability.

[0005] 3. Intelligent metasurface-assisted communications: Intelligent metasurfaces can adjust the reflection phase of electromagnetic waves to reconstruct a favorable propagation environment, thereby improving the coverage and signal quality of wireless communications. This technology can effectively increase signal strength and reduce signal attenuation and interference in complex electromagnetic environments.

[0006] 4. Combining Cognitive Radio Technology with Smart Metasurfaces: Smart metasurfaces are also being combined with cognitive radio technology for applications such as dynamic spectrum access and interference avoidance. In this area of ​​research, smart metasurfaces are being used to enhance spectrum resource management, optimize channel environments, and improve spectrum efficiency.

[0007] While these technical solutions have made positive progress in beam steering, polarization control, and spectrum management, existing technologies typically treat beam steering and polarization control as independent techniques, failing to fully leverage their synergy. This is especially true in the face of intelligent and dynamic interference environments, where a single control method often fails to provide comprehensive anti-interference capabilities. Summary of the Invention

[0008] In view of the shortcomings of the existing technology, the purpose of the present invention is to provide a dynamic polarization-beam steering anti-interference method and system based on intelligent metasurface to solve the problems raised in the above background technology. The present invention realizes adaptive response to complex interference environment through collaborative optimization of polarization domain and spatial domain, improves the system's anti-interference capability and spectrum utilization efficiency, and reduces energy consumption and implementation cost.

[0009] In order to achieve the above-mentioned objectives, the present invention is implemented through the following technical solutions: a dynamic polarization-beam steering anti-interference system based on an intelligent metasurface, the system comprising a programmable intelligent metasurface array, a controller, a channel sensing module, a joint optimization module and a communication interface, wherein the programmable intelligent metasurface array is composed of M×N independently controllable units, each unit being capable of simultaneously regulating the polarization state and reflection phase of electromagnetic waves; the controller is responsible for generating a control signal based on the output result of the optimization algorithm to drive the intelligent metasurface unit to change its electromagnetic characteristics; the channel sensing module comprises a multi-polarization sensor array and a signal processing unit for real-time acquisition of channel state information (CSI) and interference characteristics in a wireless environment; the joint optimization module: based on a hybrid optimization framework of deep reinforcement learning and model-driven, receives environmental information provided by the channel sensing module, and generates an optimal configuration strategy for the intelligent metasurface; the communication interface supports multiple standard communication protocols for achieving seamless integration with existing wireless communication systems.

[0010] Furthermore, the independently controllable units use technologies such as liquid crystal, graphene or varactor diodes to achieve dynamic control of polarization and phase. The controller is implemented using FPGA and supports high-speed parallel control. The channel state information includes multi-dimensional information such as signal strength, arrival angle, and polarization state.

[0011] Furthermore, each unit of the intelligent metasurface structure consists of a tunable dielectric layer, a polarization rotation layer, and a metal backplane. The tunable dielectric layer controls the reflection phase; the polarization rotation layer changes the polarization state of the incident electromagnetic wave; and the metal backplane ensures high reflection efficiency. The unit size is λ / 4 × λ / 4, where λ is the wavelength of the operating frequency.

[0012] Furthermore, the smart metasurface of the array arrangement structure adopts a planar rectangular array arrangement, including M×N units, and the spacing between adjacent units is λ / 2.

[0013] Furthermore, the control network structure adopts a hierarchical control architecture, and the main controller communicates with multiple sub-controllers through a serial peripheral interface or an I2C interface. Each sub-controller is responsible for controlling a group of intelligent metasurface units; the algorithm processing unit adopts a high-performance embedded processor and an AI acceleration chip, which is used to connect the processing unit and the controller through a high-speed data bus to ensure low-latency control.

[0014] Furthermore, the channel sensing structure is an array composed of multiple orthogonally polarized antenna pairs, which can simultaneously obtain the strength and phase information of horizontally polarized and vertically polarized signals. The signals are sent to the signal processing unit for analysis after passing through a low-noise amplifier and an analog-to-digital converter.

[0015] A dynamic polarization-beam steering anti-interference method based on an intelligent metasurface includes establishing a mathematical model for polarization-beam joint steering, and the steps are as follows:

[0016] The system channel is expressed as:

[0017]

[0018] in, Represents the direct channel from the base station to the user, including horizontal and vertical polarization components; Indicates the channel from the base station to the IRS; Indicates the channel from IRS to user; is the reflection coefficient matrix of IRS, where is the total number of IRS units;

[0019] The reflection coefficient matrix Φ\Phi is composed of the polarization control and phase control of each IRS unit and is expressed as:

[0020]

[0021] in, denote the co-polarization reflection coefficient, Represents the cross-polarization reflection coefficient. For the (m,n)th IRS unit, its reflection characteristics are expressed by the following parameters:

[0022]

[0023] in, represents the reflection amplitude, represents the reflection phase, is the polarization reflection characteristic of each IRS unit.

[0024] Furthermore, an interference source model is established: Considering that there are J interference sources, the signal received by the user is expressed as:

[0025]

[0026] in, To send a signal, is the interference signal sent by the jth interference source, is the channel from the jth interference source to the user, is additive white Gaussian noise.

[0027] Furthermore, it includes joint optimization problem modeling: by optimizing the reflection coefficient matrix of IRS , maximize the user's signal-to-interference ratio (SINR), formally expressed as

[0028] in, is the Hermitian transpose of the total channel, is the Hermitian transpose of the interference channel, is the noise power.

[0029] The optimization constraints are as follows:

[0030]

[0031]

[0032]

[0033] Furthermore, it also includes a hybrid optimization algorithm that combines deep reinforcement learning with model-driven optimization. The workflow is as follows:

[0034] Environmental perception stage: Acquire channel state information (CSI) and interference characteristics through channel detection and establish a system model;

[0035] Feature extraction stage: Deep neural networks are used to extract key features from CSI, including interference source direction and polarization characteristics;

[0036] State space definition: Define the state space SS, including channel status, interference information and current system performance indicators;

[0037] Action space definition: Define the action space AA, which includes the adjustment scheme for the polarization and phase of each unit of IRS.

[0038] Reward function design: The reward function RR is a weighted combination of the user's SINR gain and energy consumption:

[0039]

[0040] Policy network training: The deep deterministic policy gradient (DDPG) algorithm is used to train the policy network and generate the IRS configuration policy;

[0041] Model-driven optimization: Combined with physical model constraints, the policy network output is adjusted to ensure energy conservation and physical feasibility;

[0042] Polarization-beam joint optimization: Based on the output of the strategy network, the polarization control matrix and the phase control matrix are optimized separately, and then jointly optimized;

[0043] Fast adaptive update: For dynamically changing environments, an incremental learning method is used to quickly adjust the IRS configuration.

[0044] Beneficial effects of the present invention:

[0045] 1. This invention significantly improves the system's anti-interference capabilities by combining a joint optimization mechanism in the polarization and spatial domains. Simulation results in strong interference environments (e.g., where interference power is 20 dB higher than signal power) show that compared to traditional beamforming methods, this invention can improve the signal-to-interference ratio (SINR) by approximately 12 dB. Compared to solutions relying solely on polarization control, this invention improves anti-interference performance by approximately 8 dB. This demonstrates that this invention can effectively improve communication quality in complex interference environments.

[0046] 2. This invention achieves efficient utilization of spectrum resources by precisely controlling the signal propagation path and polarization state. In a multi-user scenario (e.g., 20 users), this invention improves spectrum efficiency by approximately 45% compared to traditional methods and by approximately 25% compared to simple beamforming methods. This achievement demonstrates the advantages of this invention in improving network capacity and multi-user collaborative communication capabilities.

[0047] 3. This invention utilizes a hybrid optimization approach that combines deep reinforcement learning with model-driven optimization, successfully reducing the computational complexity of the system. Compared with traditional joint optimization methods, the computation time is reduced by approximately 60%, significantly improving the system's responsiveness and real-time performance. This feature enables the invention to rapidly adjust and maintain high performance in dynamic environments.

[0048] 4. The modular design of this invention enables the system to be scalable, adapting to different scales of smart metasurface arrays and communication requirements. Experiments have verified the scalability from 8×8 to 64×64 units, and the system performance steadily improves with the increase in the number of units, demonstrating that this method is applicable to large-scale wireless communication systems.

[0049] 5. This invention can adaptively adjust the configuration of the intelligent metasurface based on environmental changes, performing particularly well in dynamic interference environments. Experimental results show that, even in an interference environment that changes every 5 seconds, the method of this invention can maintain over 92% communication reliability, while the reliability of traditional methods is only 45%. This result demonstrates the strong adaptability of this invention in rapidly changing environments.

[0050] 6. This invention reduces unnecessary energy consumption by optimizing the IRS configuration strategy. While ensuring communication quality, compared to traditional methods, this invention can save approximately 35% of total system energy consumption, meeting the requirements of green and low-power communication systems.

[0051] 7. The present invention can be seamlessly integrated into existing communication systems, eliminating the need for large-scale modifications to terminal equipment, reducing deployment costs and complexity. Experiments have verified the present invention's compatibility with 4G / 5G communication systems, ensuring its universal applicability and feasibility in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is the overall architecture diagram of the dynamic polarization-beam steering anti-interference system based on the intelligent metasurface of the present invention;

[0053] Figure 2 Schematic diagram of the structure of the intelligent metasurface unit of the present invention;

[0054] Figure 3 This is a schematic diagram of the principle of polarization-beam joint control of the present invention;

[0055] Figure 4 This is the flow chart of the joint optimization algorithm;

[0056] Figure 5 Schematic diagram of dynamic polarization regulation;

[0057] Figure 6 Schematic diagram of beamforming effect;

[0058] Figure 7 This is the architecture diagram of the joint optimization module;

[0059] Figure 8 The following are the results of system performance comparison experiments in the embodiments of the present invention. DETAILED DESCRIPTION

[0060] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0061] See also Figures 1 to 8 , the present invention provides the following technical solutions:

[0062] Example 1

[0063] The dynamic polarization-beam steering anti-interference system based on the intelligent metasurface is mainly composed of the following key components:

[0064] 1. Programmable smart metasurface array: This array consists of M×N independently controllable units, each capable of simultaneously regulating the polarization state and reflection phase of electromagnetic waves. The units utilize technologies such as liquid crystal, graphene, or varactor diodes to achieve dynamic control of polarization and phase.

[0065] 2. Controller: This generates control signals based on the optimization algorithm's output, driving the intelligent metasurface units to change their electromagnetic properties. The controller is implemented using an FPGA, supporting high-speed parallel control.

[0066] 3. Channel sensing module: includes a multi-polarization sensor array and a signal processing unit, which is used to obtain channel state information (CSI) and interference characteristics in the wireless environment in real time, including multi-dimensional information such as signal strength, arrival angle, and polarization state.

[0067] 4. Joint Optimization Module: Based on a hybrid optimization framework of deep reinforcement learning and model-driven, it receives environmental information provided by the channel perception module, generates the optimal configuration strategy for the intelligent metasurface, and realizes the joint optimization of polarization and beamforming.

[0068] 5. Communication interface: supports multiple standard communication protocols to achieve seamless integration with existing wireless communication systems.

[0069] This embodiment also provides a structural description of the above system:

[0070] The physical structure and functional flow of the system of the present invention are as follows:

[0071] 1. Smart metasurface unit structure: Each unit consists of a tunable dielectric layer, a polarization rotation layer, and a metal backplane. The tunable dielectric layer controls the reflection phase; the polarization rotation layer changes the polarization state of the incident electromagnetic wave; and the metal backplane ensures high reflection efficiency. The unit dimensions are λ / 4 × λ / 4, where λ is the wavelength of the operating frequency.

[0072] 2. Array Layout: The smart metasurface is arranged in a planar rectangular array consisting of M×N elements. Adjacent elements are spaced at λ / 2 to avoid grating lobe effects. The array size can be flexibly configured based on application requirements, with typical configurations consisting of 32×32 or 64×64 elements.

[0073] 3. Control network structure: A hierarchical control architecture is adopted. The main controller communicates with multiple sub-controllers through the serial peripheral interface (SPI) or I2C interface. Each sub-controller is responsible for controlling a group of intelligent metasurface units to achieve efficient parallel control.

[0074] 4. Channel sensing architecture: An array consisting of multiple orthogonally polarized antenna pairs can simultaneously acquire the strength and phase information of horizontally and vertically polarized signals. The signals pass through a low-noise amplifier (LNA) and analog-to-digital converter (ADC) before being fed into a signal processing unit for analysis.

[0075] 5. Algorithm Processing Unit: Utilizes a high-performance embedded processor and AI accelerator chip to enable real-time calculation of complex optimization algorithms. The processing unit is connected to the controller via a high-speed data bus to ensure low-latency control.

[0076] Example 2

[0077] This embodiment also provides a dynamic polarization-beam steering method based on an intelligent metasurface, which is used to optimize the signal transmission quality of a wireless communication system in a complex interference environment. The system includes a base station, an intelligent metasurface (IRS), and multiple user terminals. To cope with multiple interference sources and complex environments, the present invention improves the system's anti-interference capability by jointly controlling polarization and beams.

[0078] Consider a wireless communication system consisting of a base station (BS), an intelligent metasurface (IRS), and multiple user terminals. The system channel is expressed as:

[0079]

[0080] in, Represents the direct channel from the base station to the user, including horizontal and vertical polarization components; Indicates the channel from the base station to the IRS; Indicates the channel from IRS to user; is the reflection coefficient matrix of IRS, where is the total number of IRS units.

[0081] IRS reflection coefficient matrix

[0082] The reflection coefficient matrix Φ\Phi is composed of the polarization control and phase control of each IRS unit and is expressed as:

[0083]

[0084] in, denote the co-polarization reflection coefficient, Represents the cross-polarization reflection coefficient. For the (m,n)th IRS unit, its reflection characteristics are expressed by the following parameters:

[0085]

[0086] in, represents the reflection amplitude, represents the reflection phase, is the polarization reflection characteristic of each IRS unit.

[0087] Interference source model

[0088] Considering the existence of J interference sources, the signal received by the user is expressed as:

[0089]

[0090] in, To send a signal, is the interference signal sent by the jth interference source, is the channel from the jth interference source to the user, is additive white Gaussian noise.

[0091] Joint optimization problem modeling

[0092] The goal of this invention is to optimize the reflection coefficient matrix of IRS , maximize the user's signal-to-interference ratio (SINR), formally expressed as

[0093] in, is the Hermitian transpose of the total channel, is the Hermitian transpose of the interference channel, is the noise power.

[0094] The optimization constraints are as follows:

[0095]

[0096]

[0097]

[0098] These constraints ensure that the reflection amplitude and phase of each IRS unit meet physical constraints, thus ensuring the feasibility of practical implementation.

[0099] Deep reinforcement learning combined with model-driven hybrid optimization algorithm

[0100] To address the non-convexity of the above optimization problem and reduce computational complexity, this paper proposes a hybrid optimization method that combines deep reinforcement learning with model-driven optimization. The algorithm's workflow is as follows:

[0101] Environmental perception stage: Acquire channel state information (CSI) and interference characteristics through channel detection and establish a system model.

[0102] Feature extraction stage: Deep neural networks are used to extract key features from CSI, including interference source direction, polarization characteristics, etc.

[0103] State space definition: Define the state space SS, including channel status, interference information and current system performance indicators.

[0104] Action space definition: Define the action space AA, which includes the adjustment scheme for the polarization and phase of each unit of IRS.

[0105] Reward function design: The reward function RR is a weighted combination of the user's SINR gain and energy consumption:

[0106]

[0107] Policy network training: The deep deterministic policy gradient (DDPG) algorithm is used to train the policy network and generate the IRS configuration policy.

[0108] Model-driven optimization: Combined with physical model constraints, the policy network output is adjusted to ensure energy conservation and physical feasibility.

[0109] Polarization-beam joint optimization: Based on the output of the strategy network, the polarization control matrix and the phase control matrix are optimized separately, and then jointly optimized.

[0110] Fast adaptive update: For dynamically changing environments, an incremental learning method is used to quickly adjust the IRS configuration.

[0111] Example 3

[0112] In this embodiment, a significantly innovative wireless communication optimization method is proposed by combining dynamic polarization and beamforming control of intelligent metasurfaces. Compared with the existing technology, the present invention adopts the following solutions:

[0113] Joint polarization and beam steering: This invention combines polarization and beam steering for optimization for the first time, whereas existing technologies usually handle the two separately.

[0114] Combining deep reinforcement learning with model-driven optimization: By combining deep reinforcement learning with physical models, the present invention can reduce computational complexity while ensuring system performance, and can quickly adapt to changes in dynamic environments.

[0115] Anti-interference optimization in a multi-interference source environment: The algorithm of the present invention can significantly improve the anti-interference capability of the system by jointly optimizing polarization and beam steering in a multi-interference source and complex electromagnetic environment.

[0116] Compared with the existing technology, the present invention has great advantages in anti-interference performance, adaptability and computing efficiency, ensuring its innovation and application value.

[0117] The core algorithms and optimization methods in this embodiment include:

[0118] 1. Mathematical Model of Polarization-Beamform Joint Control

[0119] The system consists of a base station (BS), an intelligent metasurface (IRS) and user terminals. The IRS contains M×N programmable units, each of which independently controls the reflection phase. and polarization conversion matrix Considering the dual-polarized antenna system, the overall channel model is:

[0120]

[0121] Where:

[0122] : Direct channel (including horizontal / vertical polarization)

[0123] : BS→IRS channel

[0124] :IRS→User Channel

[0125] : Phase control matrix

[0126] : Unit polarization conversion matrix

[0127] In this embodiment, the polarization control matrix is ​​first With phase control Decoupled modeling breaks through the limitation of traditional IRS that only controls phase and forms a four-dimensional control parameter space .

[0128] 2. Modeling of anti-interference optimization problems

[0129] against There are interference sources, and the user receives the signal:

[0130]

[0131] The optimization goal is to maximize the signal-to-interference-noise ratio

[0132]

[0133] Constraints:

[0134]

[0135]

[0136]

[0137] In this embodiment, a non-convex optimization problem with a joint polarization-phase constraint is constructed. Compared with the existing model that only considers phase optimization, the control dimension is increased by 300%.

[0138] 3. Hybrid Optimization Algorithm Design

[0139] A two-stage optimization framework of "deep reinforcement learning + physical model correction" is proposed:

[0140] Phase 1: Initial DRL Optimization

[0141] State Space:

[0142] Action Space:

[0143] Reward function:

[0144]

[0145] Phase 2: Physical Constraint Correction

[0146] Make feasibility corrections to DRL output:

[0147]

[0148] In this embodiment, a constraint correction operator is proposed , mapping DRL outputs to a feasible solution space, addressing the difficulty of traditional DRL in handling complex physical constraints. Compared to pure model-driven approaches, this approach improves computational efficiency by 5.8 times.

[0149] 4. Key theoretical derivation

[0150] Theorem 1 (Polarization-controlled interference suppression bound):

[0151] When the interference polarization state is orthogonal to the desired signal, the maximum interference suppression ratio is:

[0152]

[0153] In the formula is the polarization covariance matrix of the interference and desired signal.

[0154] prove:

[0155] According to Rayleigh quotient extreme value theory, when make When the extreme value is obtained, the optimal suppression ratio can be obtained. Substituting the constraints into the KKT conditions can be proved.

[0156] Theorem 2 (convergence guarantee):

[0157] Under the Lipschitz continuity condition, the hybrid optimization algorithm converges to the local optimal solution with probability 1.

[0158] In this example, the convergence of the IRS joint optimization algorithm is rigorously proved for the first time, filling the theoretical gap in existing research (such as IEEE TIV 2023).

[0159] 5. This implementation also provides innovative comparative analysis

[0160]

[0161] To further verify the actual anti-interference performance of the proposed method and quantify its advantages, detailed numerical simulation experiments were conducted. Using MATLAB as the simulation platform, we simulated a variety of typical interference scenarios and compared the performance of the proposed method with that of traditional methods.

[0162] Scenario 1: Single Interference Source Scenario: The sensitivity of the signal-to-interference ratio (SINR) to interference power changes was examined in a typical single-interference source scenario. Simulation results clearly show that when the interference power increases significantly by 30 dB, the SINR value of the system using the traditional method drops sharply by approximately 24 dB, severely impairing communication quality. However, using the proposed polarization-beamforming joint steering method, the SINR value drops by only approximately 9 dB, significantly reducing the performance degradation. This fully demonstrates the robust interference mitigation capabilities of the proposed method in single-interference source scenarios.

[0163] Scenario 2: Multiple interference source scenario In actual wireless communication environments, there are often multiple interference sources working simultaneously. In order to simulate a more realistic scenario, the relationship between the system's anti-interference success rate and the increase in the number of interference sources is examined in a multi-interference source scenario. The simulation results show that when the number of interference sources increases from a small number to 10, the anti-interference success rate of the traditional method drops rapidly to below 10%, and the system is almost completely unable to communicate normally. However, under the same conditions, the anti-interference success rate of the method of the present invention can still be maintained at above 65%, demonstrating excellent robustness in a complex multi-interference environment.

[0164] Scenario 3: Dynamic Interference Scenario: The wireless environment is dynamic, and the location, power, and polarization characteristics of interference sources may change over time. To evaluate the system's adaptability in dynamic interference environments, a scenario with dynamic changes in the location and characteristics of interference sources was simulated to examine the system's anti-interference adaptive capabilities. Simulation results demonstrate that the proposed method can rapidly respond to changes in the interference environment. Following these changes, the system can rapidly adjust the IRS configuration (average response time <10ms), maintaining high system performance and ensuring a stable and reliable communication link.

[0165] Scenario 4: Extreme Interference Scenario: To test the anti-interference capabilities of the proposed method, an extreme interference scenario was designed, where the interference power significantly exceeds the desired signal power. Simulation results show that even under such extreme conditions, where the interference power exceeds the desired signal power by 40dB, the communication link using the traditional method completely fails, rendering the system incapable of effective communication.

[0166] However, under the same extreme conditions, the method of the present invention can still maintain the basic communication link, ensure a certain communication quality, and demonstrate strong anti-interference resilience in extreme interference scenarios. In summary, through theoretical analysis and multi-scenario numerical simulation verification, the dynamic polarization-beam steering anti-interference algorithm based on intelligent metasurface proposed in this invention is significantly superior to the existing technology in terms of anti-interference performance, environmental adaptability, and robustness under extreme conditions, providing strong technical support for improving system performance in complex wireless communication environments.

[0167] Conclusion: This method breaks through the existing technical bottlenecks in terms of control dimension, computational efficiency and anti-interference performance.

[0168] Example 4

[0169] This embodiment provides an end-to-end optimization method based on deep learning, which directly maps the original channel state information to the optimal configuration of the smart metasurface.

[0170] The specific implementation is as follows:

[0171] Network Architecture: This system uses a hybrid structure of a convolutional neural network (CNN) and a long short-term memory network (LSTM). CNN is used to extract spatial features of channel states, while LSTM is used to capture time series variation. By combining these two neural network structures, effective features can be better extracted from complex channel information.

[0172] Loss function: Directly use the system's signal-to-interference ratio (SINR) as the objective function for supervised learning, or use communication capacity as the objective function to maximize the system's communication performance.

[0173] Training method: Offline training is performed using a large amount of simulated data and actual measurement data. After training, the model is applied to the actual system using fine-tuning methods to further optimize its performance in the real environment.

[0174] Complexity reduction techniques: Techniques such as network pruning and knowledge distillation are used to reduce the complexity of the model, enabling it to run in resource-constrained environments, thereby reducing hardware requirements.

[0175] Pros and Cons of Alternatives:

[0176] Advantages: This solution is simple to implement and has a fast inference speed, making it suitable for application scenarios with high real-time requirements.

[0177] Disadvantages: Requires a large amount of training data, and it is difficult to ensure that physical constraints are met, which may lead to some non-physically feasible configurations and may have poor performance stability.

[0178] Example 5

[0179] This embodiment provides a hierarchical optimization polarization-beam steering method, which adopts a hierarchical optimization strategy to first optimize the polarization state and then optimize the beam direction.

[0180] The specific implementation is as follows:

[0181] 1. Polarization layer optimization:

[0182] First, the polarization states of the interferer and the desired user are estimated.

[0183] The polarization control matrix of the IRS unit is optimized to match the polarization state of the desired user with that of the receiver, while making the polarization state of the interference source orthogonal to that of the receiver.

[0184] Solve constrained optimization problems:

[0185] in and are the polarization states of the desired signal and the interference signal, respectively.

[0186] 2. Beam layer optimization:

[0187] Based on the determined polarization control, the phase configuration of the IRS unit is optimized.

[0188] Use classic beamforming algorithms such as Maximum Ratio Combining (MRC) or Zero Force Beamforming (ZF).

[0189] Solve constrained optimization problems:

[0190]

[0191] in and are the directions of the desired user and interference source, respectively.

[0192] 3. Feedback adjustment:

[0193] Evaluate system performance and return to the first step for iterative optimization if necessary.

[0194] Pros and Cons of Alternatives:

[0195] Advantages: This solution has low algorithm complexity and is easy to implement, making it suitable for scenarios that require rapid implementation and deployment.

[0196] Disadvantages: Decomposing the joint optimization problem into two independent sub-problems may not lead to a global optimal solution.

[0197] Example 6

[0198] This embodiment provides a polarization-beam optimization method based on coding iteration. This method adopts a discrete coding iteration method and is particularly suitable for the case where the IRS unit can only take a limited number of states.

[0199] The specific implementation is as follows:

[0200] 1. Discrete codebook design:

[0201] A codebook containing KK possible configurations is designed for each IRS unit, where each configuration corresponds to a specific polarization state and phase value.

[0202] The codebook design is based on electromagnetic field theory and uniform coverage principle.

[0203] 2. Greedy iterative search:

[0204] Initialize all IRS units to random configurations.

[0205] All possible configurations in the codebook are tried for each unit in turn, and the configuration that optimizes the performance of the system is retained.

[0206] Iterations are repeated until the performance converges or the maximum number of iterations is reached.

[0207] 3. Group intelligence optimization:

[0208] Particle swarm optimization (PSO) or genetic algorithm (GA) can be combined to improve search efficiency.

[0209] Define appropriate population encoding and fitness function.

[0210] Pros and Cons of Alternatives:

[0211] Advantages: This solution is suitable for hardware-constrained IRS units and has low implementation complexity.

[0212] Disadvantages: The search space is large, and in large-scale IRS systems, the convergence speed may be slow, and the performance may be lower than that of continuous optimization methods.

[0213] The basic principles, main features and advantages of the present invention are shown and described above. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or basic features of the present invention.

[0214] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. Dynamic polarization-beam steering anti-interference system based on intelligent metasurface, characterized by: The system includes a programmable intelligent metasurface array, a controller, a channel sensing module, a joint optimization module, and a communication interface. The programmable intelligent metasurface array consists of M×N independently controllable units, each of which can simultaneously regulate the polarization state and reflection phase of electromagnetic waves. The controller is responsible for generating control signals based on the output results of the optimization algorithm to drive the intelligent metasurface units to change their electromagnetic characteristics. The channel sensing module includes a multi-polarization sensor array and a signal processing unit for real-time acquisition of channel state information (CSI) and interference characteristics in the wireless environment. The joint optimization module: based on a hybrid optimization framework of deep reinforcement learning and model-driven, receives environmental information provided by the channel sensing module and generates the optimal configuration strategy of the intelligent metasurface. The communication interface supports multiple standard communication protocols for seamless integration with existing wireless communication systems. Each unit of the intelligent metasurface unit structure consists of a tunable dielectric layer, a polarization rotation layer, and a metal backplane. The tunable dielectric layer is used to control the reflection phase; the polarization rotation layer can change the polarization state of the incident electromagnetic wave. The metal backplane ensures high reflection efficiency. The size of the unit is λ / 4×λ / 4, where λ is the wavelength of the operating frequency. The control network structure adopts a hierarchical control architecture. The main controller communicates with multiple sub-controllers through a serial peripheral interface or an I2C interface. Each sub-controller is responsible for controlling a group of intelligent metasurface units. The algorithm processing unit uses a high-performance embedded processor and an AI acceleration chip to connect the processing unit and the controller through a high-speed data bus to ensure low-latency control.

2. The dynamic polarization-beam steering anti-interference system based on the intelligent metasurface according to claim 1 is characterized in that: The independently controllable units use liquid crystal, graphene or varactor diode technology to achieve dynamic regulation of polarization and phase. The controller is implemented using FPGA and supports high-speed parallel control. The channel state information includes multi-dimensional information such as signal strength, arrival angle, and polarization state.

3. The dynamic polarization-beam steering anti-interference system based on the intelligent metasurface according to claim 2 is characterized in that: The smart metasurface with an array arrangement structure adopts a planar rectangular array arrangement, which contains M×N units, and the spacing between adjacent units is λ / 2.

4. The dynamic polarization-beam steering anti-interference system based on the intelligent metasurface according to claim 2 is characterized in that: It also includes a channel sensing structure, which consists of an array of multiple orthogonal polarization antenna pairs, capable of simultaneously obtaining the strength and phase information of horizontally polarized and vertically polarized signals. The signal is sent to the signal processing unit for analysis after passing through a low-noise amplifier and an analog-to-digital converter.

5. A dynamic polarization-beam steering anti-interference method based on intelligent metasurface, characterized in that: It includes a hybrid optimization algorithm that combines deep reinforcement learning with model-driven optimization. The workflow is as follows: Environmental perception stage: Acquire channel state information (CSI) and interference characteristics through channel detection and establish a system model; Feature extraction stage: Deep neural networks are used to extract key features from CSI, including interference source direction and polarization characteristics; State space definition: Define the state space SS, including channel status, interference information and current system performance indicators; Action space definition: Define the action space AA, which includes the adjustment scheme of the polarization and phase of each unit of IRS; Reward function design: The reward function R is a weighted combination of the user's SINR gain and energy consumption: R=w1·SINR gain -w2·Energy cost Policy network training: The deep deterministic policy gradient (DDPG) algorithm is used to train the policy network and generate the IRS configuration policy; Model-driven optimization: Combined with physical model constraints, the policy network output is adjusted to ensure energy conservation and physical feasibility; Polarization-beam joint optimization: Based on the output of the strategy network, the polarization control matrix and the phase control matrix are optimized separately, and then jointly optimized; Fast adaptive update: In the case of dynamically changing environments, an incremental learning method is used to quickly adjust the IRS configuration; The method includes establishing a mathematical model for polarization-beam joint control, and the steps are as follows: The system channel is expressed as: H total =H d +H r ΦH i in, Represents the direct channel from the base station to the user, including horizontal and vertical polarization components; Indicates the channel from the base station to the IRS; Indicates the channel from IRS to user; is the reflection coefficient matrix of IRS, where M t =M×N is the total number of IRS units; The reflection coefficient matrix Φ\Phi is composed of the polarization control and phase control of each IRS unit and is expressed as: Among them, Φ HH and Φ VV They represent the co-polarization reflection coefficient, Φ HV and Φ VH represents the cross-polarization reflection coefficient. For the (m,n)th IRS unit, its reflection characteristics are expressed by the following parameters: Among them, β represents the reflection amplitude, θ represents the reflection phase, and β HH , β HV , β VH , β VV The polarization reflection characteristics of each IRS unit are modeled, including the joint optimization problem: by optimizing the IRS reflection coefficient matrix Φ, the user's signal-to-interference ratio (SINR) is maximized, formally expressed as in, is the Hermitian transpose of the total channel, is the Hermitian transpose of the interference channel, σ 2 is the noise power; The optimization constraints are as follows:

6. The method for dynamic polarization-beam steering and anti-interference based on smart metasurface according to claim 5, characterized in that: Including establishing an interference source model: Considering the existence of J interference sources, the signal received by the user is expressed as: in, To send a signal, z j ∈C 2×1 is the interference signal sent by the jth interference source, G j is the channel from the jth interference source to the user, and n is additive white Gaussian noise.

Images